Technological advances and affordability of recent smart sensors, as well as the consolidation of common software platforms for the integration of the latter and robotic sensors, are enabling the creation of complex active and assisted living environments for improving the quality of life of the elderly and the less able people. One such example is the integrated system developed by the European project ENRICHME, the aim of which is to monitor and prolong the independent living of old people affected by mild cognitive impairments with a combination of smart-home, robotics and web technologies. This paper presents in particular the design and technological solutions adopted to integrate, process and store the information provided by a set of fixed smart sensors and mobile robot sensors in a domestic scenario, including presence and contact detectors, environmental sensors, and RFID-tagged objects, for long-term user monitoring and adaptation.
Wind power, as an alternative to burning fossil fuels, is plentiful and renewable. Data-driven approaches are increasingly popular for inspecting the wind turbine failures. In this paper, we propose a novel classification-based anomaly detection system for icing detection of the wind turbine blades. We effectively combine the deep neural networks and wavelet transformation to identify such failures sequentially across the time. In the training phase, we present a wavelet based fully convolutional neural network (FCNN), namely WaveletFCNN, for the time series classification. We improve the original (FCNN) by augmenting features with the wavelet coefficients. WaveletFCNN outperforms the state-of-the-art FCNN for the univariate time series classification on the UCR time series archive benchmarks. In the detecting phase, we combine the sliding window and majority vote algorithms to provide the timely monitoring of the anomalies. The system has been successfully implemented on a real-world dataset from Goldwind Inc, where the classifier is trained on a multivariate time series dataset and the monitoring algorithm is implemented to capture the abnormal condition on signals from a wind farm.
Variational Autoencoders (VAE) are probabilistic deep generative models underpinned by elegant theory, stable training processes, and meaningful manifold representations. However, they produce blurry images due to a lack of explicit emphasis over high-frequency textural details of the images, and the difficulty to directly model the complex joint probability distribution over the high-dimensional image space. In this work, we approach these two challenges with a novel wavelet space VAE that uses the decoder to model the images in the wavelet coefficient space. This enables the VAE to emphasize over high-frequency components within an image obtained via wavelet decomposition. Additionally, by decomposing the complex function of generating high-dimensional images into inverse wavelet transformation and generation of wavelet coefficients, the latter becomes simpler to model by the VAE. We empirically validate that deep generative models operating in the wavelet space can generate images of higher quality than the image (RGB) space counterparts. Quantitatively, on benchmark natural image datasets, we achieve consistently better FID scores than VAE based architectures and competitive FID scores with a variety of GAN models for the same architectural and experimental setup. Furthermore, the proposed wavelet-based generative model retains desirable attributes like disentangled and informative latent representation without losing the quality in the generated samples.
Tremblay, Sebastien (Universite du Quebec a Chicoutimi (UQAC)) | Fortin-Simard, Dany (Universite du Quebec a Chicoutimi (UQAC)) | Blackburn-Verreault, Erika (Universite du Quebec a Chicoutimi (UQAC)) | Gaboury, Sebastien (Universite du Quebec a Chicoutimi (UQAC)) | Bouchard, Bruno (Universite du Quebec a Chicoutimi (UQAC)) | Bouzouane, Abdenour (Universite du Quebec a Chicoutimi (UQAC))
The number of elderly and frail individuals in need of daily assistance increases and the available human resources will certainly be insufficient. To remedy this situation, smart habitats are considered by many researchers as an innovative avenue to help support the needs of elders. It aims at providing cognitive assistance in taking decisions by giving hints, suggestions, and reminders with different kinds of effectors to residents. To implement such technology, the first challenge we need to overcome is the recognition of the ongoing activity. In the literature, some researchers have proposed solutions based on cameras, binary sensors, radio-frequency identification and load signatures of appliances but all these types of approaches have certain limitations to perform a complete recognition. In order to provide additional and useful information, a complementary activity recognition system, based on environmental sounds and able to detect errors related to cognitive impairment, is presented in this paper. The entire system relies on a discrete wavelet transform, the zero-crossing rate and C4.5 algorithm. This system has been implemented and deployed in a real smart-home prototype. This paper also present the results of a first set of experiments conducted on this system with real cases scenarios.
Sophisticated malware authors can sneak hidden malicious code into portable executable files, and this code can be hard to detect, especially if it is encrypted or compressed. However, when an executable file shifts between native code, encrypted or compressed code, and padding, there are corresponding shifts in the file's representation as an entropy signal. In this paper, we develop a method for automatically quantifying the extent to which the patterned variations in a file's entropy signal makes it "suspicious." A corpus of n = 39,968 portable executable files were studied, 50% of which were malicious. Each portable executable file was represented as an entropy stream, where each value in the entropy stream describes the amount of entropy at a particular locations in the file. Wavelet transforms were then applied to this entropy signal in order to extract the amount of entropic energy at multiple scales of code resolution. Based on this entropic energy spectrum, we derive a Suspiciously Structured Entropic Change Score (SSECS), a single scalar feature which quantifies the extent to which a given file's entropic energy spectrum makes the file suspicious as possible malware. We found that, based on SSECS alone, it was possible to predict with 68.7% accuracy whether a file in this corpus was malicious or legitimate (a 18.7% gain over random guessing). Moreover, we found that SSECS contains predictive information not contained in mean entropy alone. Thus, we argue that SSECS could be a useful single feature for machine learning models which attempt to identify malware based on millions of file features.